Vineyards are predominantly located in drought prone areas. They commonly experience moderate soil water deficit, which is favorable to wine quality provided that it remains moderate (Becker and Zimmerman, 1984). Excess of water, by contrast, can reduce color intensity and sugar content of berries and produce unbalanced, flat wine (Matthews et al., 1990; Medrano et al., 2003). Thus, moderate soil water deficit is the best compromise to promote the expression of high enological potential without altering yield. This is usually achieved in most vineyards but global change seriously threatens this fragile equilibrium. Specifically, under the combined influence of high evaporative demand (dry, warm air) and soil water deficit, plant tissues start dehydrating with detrimental impacts on production and berry quality (Jones et al., 2005; Deluc et al., 2009).

To face transient drought or longer-lasting dry climates, irrigation is developing in production areas. However, pressure on agricultural use of water resources is rising. Irrigation of the vineyard often results as very competitive or impossible. To prepare for the future, viticulture should adapt by limiting water use while maintaining yield. Vineyard establishment and management practices, such as lower plantation density, control of water balance through soil surface management, and thinning, can be considered as valuable short-term solutions (Garcia de Cortazar Atauri, 2006; Duchêne et al., 2010; Ripoche et al., 2010). However, these techniques might not always be sufficient to cope with increasingly dryer conditions (Garcia de Cortazar Atauri, 2006). Additional strategies are needed, including the use of suitable plant material. This requires a comprehensive knowledge of the physiological impacts of drought on yield and quality.

In the following, we review the primary consequences of water deficit on grapevine. Specifically, genetic variability in the mechanisms involved in the control of plant water status is examined.

Physiological responses to water deficit

Drop in plant water potential as a primary consequence of water deficit

Water potential characterizes water availability from a thermodynamic point of view. Denoted Ψ, it is at the basis of water movements from the soil to the plant organs and ultimately to the atmosphere. Conventionally, free water at sea level has a potential of zero, corresponding to the maximal water availability in a saturated soil. Soil drying results in a decrease of soil water potential (Ψ becomes more negative as water binds to soil particles and concentrating solutes). Under non transpiring conditions, water potentials in plants equilibrate with the most humid layer explored by the root system (Améglio et al., 1999). As transpiration rate increases in the daytime, plant water potential decreases. This drop in water potential is more severe when hydraulic conductance is limiting water transport on the path from the soil through the plant to the leaves (Figure 1). Because excessive drops in water potential may be disastrous for plants, they have developed diverse adaptations to prevent them.

Cavitation threatens hydraulic integrity of xylem conduits

In a transpiring plant, sap water ascends towards the leaves using the non-living, heavily thickened and lignified xylem vessels and tracheids. Water flow follows a gradient of increasingly negative pressure within a continuous water column. Any break in this column would disrupt the whole water flow.

When soil drying combines with high evaporative demand, high tensile strength develops in the xylem, thereby favoring cavitation, which is the apparition of gaseous bubbles (caveats) in the xylem sap due to water evaporation, aggregation of dissolved gases or air entry through pit membranes. Once initiated the bubble then rapidly expands to overrun the vessel (Brodersen et al., 2013). This gaseous embolism may result in the rupture of the water column in the xylem, being a major threat for the plant.

Vessel embolism decreases stem hydraulic conductance, which in turn decreases leaf water potential itself, favoring further embolism. In the absence of stomatal closure or reduction in leaf area, this cycle can result in functionality loss of all the conducting tissue. This results in dramatically amplified effects of water deficit on the drop in leaf water potential along the water path (Brodribb and Cochard, 2009; Zufferey et al., 2011) with catastrophic consequences on plant dehydration and even death (McDowell et al., 2008). Vessel size partly determines plant vulnerability to cavitation, small-diameter conduits being less vulnerable (Tyree, 2003) but less efficient to transport water. Thus, plant adaptation to dry environments depends on a trade-off between efficient conduits and low vulnerability to cavitation. In grapevine, which displays long vessels (a common feature among liana species), vessel sizes are dependent on the cultivar (Chouzouri and Schultz, 2005; Tramontini et al., 2013a), leaving room for genetic variation in drought response.

Threshold water potential for cavitation also varies with species, cultivars and growth conditions. As compared to other species, grapevine has commonly been described as vulnerable to cavitation occurring at high (less negative) water potential threshold (Schultz and Matthews, 1988; McElrone et al., 2012). Up to 70% loss of conductivity has been reported with moderate tensions in stems around -0.75 MPa (Tibbetts and Ewers, 2000). Nevertheless, an efficient control of water losses through stomata often protects grapevine from cavitation (Zufferey et al., 2011).

Recent studies report that transport capacity could be largely restored by the end of the day or during the night, when transpiration rate decreases. This has been assigned to water refilling of embolized xylem vessels. Although mechanistically debated, restoration of water transport capacity has been observed in a number of species, whether in roots (Domec et al., 2006; Lovisolo et al., 2008a), shoots (Zwieniecki and Holbrook, 1998) or leaves (Johnson et al., 2009). Plant capacity to restore hydraulic integrity over night under dry conditions would largely depend on soil exploration by roots (Zufferey et al., 2011). Carbohydrates stored in cells neighboring the conducting vessels, together with aquaporins (e.g. membrane channel proteins facilitating water transport), also appear as possible, important actors of this restoration (Salleo et al., 2009).

Limitation of transpiration releases hydraulic tension and saves water

One of the most obvious and immediate effects of water deficit is a reduction in shoot growth (Chaves, 1991), with cell expansion being particularly sensitive to water shortage (Hsiao, 1973). Branches are more sensitive than first order axes (Lebon et al., 2006), and observation of growth cessation at the shoot apices is a powerful tool to early detect incipient water deficit (Pellegrino et al., 2006). Limited vigor under drought results in a decrease of evaporative areas, thereby lowering transpiration and releasing water tension in the xylem. Leaf folding or wilting are other adaptations having similar, although reversible, effects on water saving by increasing boundary layer resistance and reducing intercepted light, hence lowering surface temperature and evaporative demand.

Figure . Physiological responses associated with a drop in plant water potential. Simplified representation adapted from Chaves et al. (2010) and Marguerit (2010). The left diagram illustrates the drop of water potential occurring at different intensities depending on the soil water potential (dryer from right to left) and of the evaporative demand (higher from right to left). During the night, water potential equilibrates (vertical line). In the daytime, under high evaporative demand, plant water potentials become more negative (dashed lines) and further decline in dry soil (dotted line). Arrows indicate the influence of the physiological adaptations (limitation of water losses, maintenance of hydraulic conductance and enhanced water uptake) on water potentials, highlighting the favorable (filled, black arrows) and the unfavorable situations (dotted, grey arrows). The right diagram outlines the main physiological adaptations favoring the maintenance of plant water status. The negative consequences of a decrease in leaf water potential on carbon assimilation are also highlighted.

Additionally, plants dynamically modulate the aperture of stomata, those micropores located at the leaf surface that make possible water vapor and CO2 exchanges. A rapid stomatal closure is generally observed under water deficit (Damour et al., 2010), which efficiently lowers water flow density. However, this way of saving water has a heavy cost for the plant because stomatal closure unavoidably lowers CO2 uptake and decreases photosynthesis, although to variable extent depending on species and varieties (Tardieu and Simonneau, 1998). Plants thus face a dilemma, and adaptive strategies are necessary to reach a trade-off ensuring CO2 uptake while limiting water losses.

Other adaptations may participate in minimizing transpiration rate, including changes in thickness and composition of the waxy cuticle that waterproofs the leaf surface and forces water to leave the plant through stomata. Relation between cuticle components and their efficacy to limit water losses remains to be understood (Riederer and Schreiber, 2001).

High root water extraction capacity postpones the negative impact of water deficit

Root development is highly plastic, with typical shifts in the allocation of plant’s resources (carbohydrates) towards root growth at the expense of the shoots in dry conditions. This allows the plant to increase soil exploration for water uptake while reducing transpiration (Sharp and Davies, 1985; Cramer et al., 2013). The maintenance of root growth capacities during water deficit, together with some plasticity in root hydraulic architecture under fluctuating conditions, depend on the species and, in grapevine, is variable among rootstocks (Bauerle et al., 2008).

Osmotic adjustment helps maintaining water into the cells

Plants evolved in different ways to maintain physiological activity while water potential declines. A major response is osmotic adjustment, which allows the cells to maintain their water content and turgor even when water potential decreases in their vicinity. Osmotic adjustment in a cell consists of trapping or generating solutes to increase their concentration, leading to interactions of water with solutes inside the cell. This decreases the osmotic potential, a component of the total water potential, while turgor, the other component in cells, can be maintained even when a given drop in total water potential is transmitted to the cell from its environment.

This widespread response to water stress occurs in leaves, roots and reproductive organs of many species (Turner and Jones, 1980; Morgan, 1984) and is under genetic control (e.g. Teulat et al., 2001). In grapevine, osmotic adjustment has been evidenced under water deficit in leaves (Rodrigues et al., 1993) and roots (During and Dry, 1995). It might be a major strategy to avoid tissue dehydration and maintain grapevine production in dry conditions (Hare et al., 1998; Patakas and Noitsakis, 1999). The most interesting solutes are those that, besides their role in osmotic adjustment, play a role in nutrient or energy storage, membrane protection or detoxifying activities (Szabados et al., 2011).

Primary traits for a drought tolerant grapevine ideotype

Plant responses to drought are plural and involve a range of morphological and physiological adaptations of both aerial and underground organs. The primary features of interest for grapevine encompass a tight control of water losses through stomatal regulation, osmoregulation, together with photosynthesis maintenance to the benefit of berry development and root growth. The tight coupling between photosynthesis and transpiration, which are both controlled by stomata and leaf area, does not make trivial to decrease transpiration without altering photosynthesis. However, the ratio of photosynthesis to transpiration rates varies to some extent with enviroental conditions and genotypes (Tomás et al., 2014; Medrano et al., 2015). An adequate control of stomatal aperture allows the plant to take advantage of the environmental conditions by lowering the water cost of gas exchange.

Physiological control of leaf water potential in a drying soil

The stomatal control of transpiration

Transpirational water losses, which, in combination with soil drying, are responsible for drawing down water potential in plants, mainly occur through the stomata. Stomata form microscopic pores mainly located on the abaxial (inferior) epidermis of the leaves in grapevine, a species therefore qualified as hypostomatous. A pair of adjacent guard cells controls the pore aperture through rapid modification in cell volume associated with turgor changes. Changes in turgor result either from variations in total water potential driven by soil or air drying (hydraulic response), or from active changes in osmotic potential caused by solute movements (into or out of the guard cells), themselves generated by chemical signals that modify ion transporter activity (biochemical response). Moreover, stomatal density displays a high inter-specific and intra-specific variability, as exemplified for grapevine (Boso et al., 2011). However, variability in stomatal density was not found to explain much of the differences in transpiration rate (Hopper et al., 2014).

Stomatal closure in response to water deficit is controlled by abscisic acid (ABA), a plant hormone having long been recognized as a key player in plant abiotic stress responses (Loveys, 1984; Wilkinson and Davies, 2002; Yamaguchi-Shinozaki and Shinozaki, 2006). ABA biosynthesis, metabolism, and transfer towards guard cells modulate stomatal sensitivity to water deficit (Stoll et al., 2000; Cramer et al., 2007). ABA synthesis in roots was first proposed as the pivot of plant response to drought. Soil drying is sensed by the roots as their water potential decreases, resulting in an increased ABA biosynthesis by this compartment (Simonneau et al., 1998). ABA is then conveyed to the leaves through the xylem vessels (Tardieu and Simonneau, 1998). ABA biosynthesis also occurs in the leaves (Holbrook et al., 2002; Christmann et al., 2005; Christmann et al., 2007; Ikegami et al., 2009) where hydraulic and chemical signals trigger foliar ABA synthesis in response to water deficit (Christmann et al., 2013; Mittler and Blumwald, 2015), although the precise signal transduction still remains to be deciphered. Several key enzymes of the ABA biosynthetic pathway, namely ABA2, AAO3, and NCED3, are expressed in specific areas of vascular tissues in response to water deficit (Endo et al., 2008). Importantly, VvNCED1 coding for 9-cis-epoxycarotenoïd dioxygenase NCED, an enzyme catalyzing the first committed step in ABA biosynthesis, has been identified as decisive for ABA accumulation under water shortage in grapevine (Speirs et al., 2013; Rossdeutsch et al., 2016). Variations of pH between tissues, together with the action of glucosidases or glucosyl esterases, modify the concentration of free ABA reaching the stomata (Nambara and Marion-Poll, 2005). Depletion of ABA may also participate in the regulation of ABA balance. A specific group of enzymes, including the ABA 8’-hydroxylases, regulates ABA degradation to inactive compounds (Speirs et al., 2013). A strong allelic diversity for genes involved in either ABA biosynthesis or degradation could explain genetic variations in ABA accumulation under water deficit (Nambara and Marion-Poll, 2005; Riahi et al., 2013). In grapevine, variability in ABA accumulation has been observed among rootstocks (Peccoux, 2011) as well as scions (Soar et al., 2004).

Additionally to ABA accumulation, stomatal sensitivity to the hormone is also highly variable (Tardieu and Simonneau, 1998; Rossdeutsch et al., 2016). It depends on numerous molecular steps at the guard cell level. Perception of ABA corresponds to binding to the PYR/PYL/RCAR proteins (Brandt et al., 2012). This leads to conformational change in the receptor enabling ABA interaction with PP2Cs phosphatase, which in turn releases SnRK2s kinases. SnRK2s activate transcription factors, ABA-responsive element Binding Factors (ABFs), which results in ABA-responsive gene expression (Klingler et al., 2010; Boneh et al., 2012). This cascade modulates the activity of ion channels in the guard cells, which translates in osmotic and turgor changes, and ultimately regulates stomatal closure (Joshi-Saha et al., 2011). Many other actors involved in those responses have been identified, including variations in internal Ca2+ concentration and accumulation of nitrous oxide in guard cells.

How chemical control of stomatal aperture interacts with hydraulics is still a matter of debate. It has recently been proposed that ABA might affect leaf hydraulic conductance through a decrease in water permeability within leaf vascular tissues. ABA would thus promote stomatal closure in a dual way via effects on hydraulics upstream stomata and a direct biochemical effect on the guard cells (Pantin et al., 2013). Variability in the role of ABA on hydraulic conductance remains to be explored as a possible cause of the large diversity of stomatal sensitivities to ABA observed among species and within grapevine cultivars.

Isohydric genotypes are able to maintain leaf water potential in drying soils

Soil drying inevitably results in a decrease of water potential in plants including leaves. However, contrasting controls of leaf water potential have been observed across species when submitted to similar soil water deficit conditions (Tardieu and Simonneau, 1998). So-called isohydric species, such as maize, efficiently maintain high leaf water potential in the daytime (ΨM) when the soil dries, whereas anisohydric species, such as sunflower, exhibit substantial decrease of ΨM (Tardieu et al., 1996). In several species including the overall, roughly isohydric grapevine (Prieto et al., 2010), a variable efficacy to maintain high ΨM has been observed across genotypes. Two widespread cultivars, namely Grenache and Syrah, have been consistently described with different responses to soil water deficit. Grenache was shown to be near-isohydric, compared with Syrah, which exhibited more anisohydric behavior (Schultz, 2003; Soar et al., 2006b).

The classical view relates the contrasted (an)isohydric behaviors to the more or less efficient control of transpiration rate by stomatal closure (Buckley, 2005). Stomatal conductance was shown to decrease earlier during the course of a soil drying episode in isohydric species, thus reducing the drop of leaf water potential in the daytime as compared to anisohydric species (Tardieu and Simonneau, 1998). The anisohydric behavior would thus favor photosynthesis maintenance under water deficit. This has been confirmed in grapevine (Lovisolo et al., 2010) where anisohydric cultivars also exhibit higher vigor in conditions of water deficit (Pou et al., 2012), as long as soil drying does not induce any serious decrease of plant water potential. Anisohydric plants might also be more resistant to cavitation than isohydric ones (Schultz, 2003; Alsina et al., 2007) and might easily recover from partial cavitation events, thus exhibiting a higher tolerance to moderate water deficit events. However, beyond a certain threshold in soil drying, the anisohydric behavior might not remain favorable because high levels of dehydration lead to serious damages. This has been exemplified for grapevine cultivars such as Syrah and Chardonnay (Alsina et al., 2007). By contrast, the isohydric cultivar Cabernet-Sauvignon displays a reduced photosynthesis but is preserved against damages such as photoinhibition, which is the alteration of photosynthesis due to high light intensity (Hochberg et al., 2013). Hence, one of these behaviors can be more interesting depending on the water deficit scenario (duration, intensity, combination with evaporative demand). While anisohydric cultivars may be recommended in the case of short periods of moderate water deficit because they sustain production, the isohydric ones appear as more suitable to face long lasting periods of severe drought. Specificities of the climatic scenarios should be considered to define the more advantageous type of cultivar from an agronomic point of view.

Reconsidering the origin of the variation in (an)isohydric behaviors

The classical view of (an)isohydry was recently questioned in several studies. It was proposed that changes in hydraulic conductance may contribute, concurrently with stomatal regulation, to the control of ΨM under adverse conditions (Franks et al., 2007; Pantin et al., 2013). Additionally, (an)isohydry would not be a genotype-constitutive feature (Lovisolo et al., 2010) but could vary in a same plant following season and development (Poni et al., 1993; Chaves et al., 2010). Some studies concluded to variable ranking of (an)isohydric behaviors between grapevine cultivars, notably Grenache and Syrah (Pou et al., 2012). The genetic origin of (an)isohydry was thus challenged.

Genetic variation in (an)isohydry was extensively studied in grapevine using a mapping population obtained from a cross between Syrah and Grenache (Coupel-Ledru et al., 2014). Significant genetic control of ΨM under moderate drought was observed under controlled conditions using potted plants in a phenotyping platform. Several genomic regions (QTLs) were identified as underlying the genetic variation of ΨM. Further, the maintenance of ΨM under water deficit conditions was not simply controlled by transpiration response to soil drought. Some of the QTLs detected for genetic variation in ΨM response to moderate water deficit collocated with QTLs for transpiration response, but others collocated with QTLs detected for plant hydraulic conductance (Coupel-Ledru et al., 2014). Overall, genetic variation of ΨM under water deficit conditions correlated with variation in plant hydraulic conductance (Coupel-Ledru, 2015). It was thus proposed that whole plant hydraulic conductance under water deficit might combine with stomatal control of transpiration to determine (an)isohydry. Specifically, variation in (an)isohydry may result from slight deviation in the balance between transpiration rate and hydraulic conductance.

The genetic analysis of the Syrah × Grenache offspring (Coupel-Ledru et al., 2014) also evidenced that transpiration rate and soil-to-leaf hydraulic conductance mostly correlated. This may explain why grapevine can be considered as roughly isohydric by contrast with other species like sunflower where more severe drops in ΨM rapidly occur as the soil dries (Tardieu et al., 1996). In grapevine, this balance may be the result of multiple coordination between stomatal response and variation in specific hydraulic conductance in leaves (Pou et al., 2012), petioles (Schultz, 2003) and roots where correlation with expressions of water channel proteins in roots has been evidenced (Vandeleur et al., 2009). Identification of genes specifically associated with QTLs detected for hydraulic conductance and control of ΨM but not for transpiration response (and vice versa) would be of particular interest to look for origins of possible imbalance between transpiration and water transport capacity and to progress on the determinism of (an)isohydry.

Rootstocks: the hidden half

While the choice of scion varieties is often regulated by their performance in specific climatic conditions or marketing purposes (van Leeuwen and Seguin, 2006), rootstocks offer more flexible solutions for adapting the grafted plant to drought. A large variability in rootstock response to water deficit has been reported by several authors (Carbonneau, 1985; Ollat et al., 2016; Zhang et al., 2016), although underlying mechanisms still need to be enlightened. Rootstocks participate in the regulation of plant water balance through their own uptake capacities associated with root growth and water transport (Carbonneau, 1985; Bauerle et al., 2008; Alsina et al., 2011; Peccoux, 2011; Zhang et al., 2016) or via their effects on stomatal regulation (Lovisolo et al., 2010; Marguerit et al., 2012) and above ground development (Jones, 2012). Water extraction capacities by roots are reported to be variable between rootstocks and genetically controlled (Carbonneau, 1985; Soar et al., 2006a; Marguerit et al., 2012), even though the physiological mechanisms underlying this trait are still unknown. In addition, rootstocks are known to affect scion phenology, vegetative growth, yield and fruit quality (Tandonnet et al., 2010).

Root development to better explore soil water resources

A deep and dense root system favors water uptake to compensate for water losses by transpiration. Grapevine is known for its ability to grow deep roots. Root distribution and root system architecture are more affected by soil type and training system than by rootstock genotype (Smart et al., 2006). In addition, interactions with scion genotypes have a strong effect on root system development (Tandonnet et al., 2010). By contrast, rootstock genotype has more impact on root density expressed as biomass - or root number by volume of soil - (Southey and Archer, 1988; Peccoux, 2011), or on the ratio of fine roots to total roots (Van Zyl, 1988). In the vineyard, some highly drought tolerant rootstocks such as 140Ru are more able to grow roots in deep soil layers (Southey and Archer, 1988). Furthermore, the maintenance of root growth under dry conditions as well as the root system plasticity with soil water status may differentiate rootstock genotypes (Bauerle et al., 2008) according to their strategy to cope with drought (Comas et al., 2010). Further investigations of root growth properties for different rootstocks would be profitable for the future.

The control of water transport to shoot

The root system contributes in a non-negligible way to the whole plant resistance to water flow (Steudle, 2000). There is a large variability among rootstocks in root vascular anatomy (vessel diameter and length, percentage of conducting tissues; Pongracz and Beukman, 1970; Alsina et al., 2011; Peccoux, 2011). These differences can affect root ability to convey water to the canopy (i.e. hydraulic conductance), and root vulnerability to cavitation. Differences between rootstocks for root hydraulic conductance have been reported (de Herralde et al., 2006; Peccoux, 2011; Tramontini et al., 2013b), but may be more related to whole root system size than individual root properties (Alsina et al., 2011). In addition, drought effect on root hydraulic conductivity may differ between rootstocks. Barrios-Masias et al. (2015) observed a lower decrease of root conductivity for the drought tolerant rootstock 110R, in comparison to the drought sensitive 101-14MGt. Differences are related to the development of suberized apoplastic barriers in the root tips at the beginning of the maturation zone. In grapevine, roots together with leaves are more sensitive to embolism than the other plant compartments (Tramontini and Lovisolo, 2016). Besides, it was recently shown that wild Vitis species stems differ for their sensitivity to cavitation under water stress and their ability to repair after rehydration, paralleling contrasting responses of root pressure to re-watering associated to osmotic regulation (Knipfer et al., 2015).

Without any doubt, these facts have specific, molecular origins in the context of grafted plants. Transcriptomic analyses in the root tissues of various rootstock-scion combinations submitted to long term water deficit support the involvement of cell wall and osmotic metabolisms in the variability of responses among rootstocks (Peccoux, 2011).

Aquaporins as key actors in transmembrane water transport

The ability to drive water from root tips to stomata does not only depend on vascular pathways. Water also follows inter- and intracellular pathways that are under the control of water channel proteins embedded in cell membranes, named aquaporins (Maurel et al., 2015). The genes encoding for such proteins have been identified for grapevine (Fouquet et al., 2008; Shelden et al., 2009). Their expression has been reported in different plant compartments, various genotypes and under drought conditions (Galmes et al., 2007; Gambetta et al., 2012; Rossdeutsch, 2015). Some of these genes are more expressed in root tips than in more mature suberized zones of the roots where the radial hydraulic conductivity is lower (Gambetta et al., 2013). Differences have been reported among rootstock genotypes for the expression of these genes under well-watered and drought conditions, or for the proportion of conductance under the control of aquaporins (Lovisolo et al., 2008b; Gambetta et al., 2012; Rossdeutsch, 2015). Although the expression of some aquaporin genes like VvPIP1;1 in roots appears to correlate with hydraulic conductance and plant transpiration (Vandeleur et al., 2009), the situation in a grafted plant is much more complex and scion effects have to be considered as well (Tramontini et al., 2013b; Rossdeutsch, 2015). Rootstocks also present contrasting abilities to produce ABA under drought conditions (Rossdeutsch et al., 2016) and the interactions of chemical and hydraulic signals of soil water status from rootstock to scion should be taken into account. The role of ABA in the control of the expression and activity of aquaporins is now clearly established (Finkelstein, 2013; Grondin et al., 2015).

Genetic architecture for transpiration and growth as controlled by rootstock

The genetic architecture for water deficit responses induced by rootstock remains poorly studied (Marguerit, 2010; Marguerit et al., 2012). Specifically, it can be questioned to what extent transpiration, growth and water use efficiency are genetically controlled by the rootstock. This question has been addressed in a 3-year experiment using a pedigree population issued from the cross between V. vinifera Cabernet Sauvignon × V. riparia Gloire de Montpellier made up of 138 individuals. Transpiration rate, δ13C (a proxy for water use efficiency), transpiration efficiency (ratio of biomass produced to water transpired), water extraction capacity and the response of transpiration to water deficit were characterized. Broad sense heritability was above 0.3 for most traits, although with significant year effects highlighting the strong impact of the environment. Few significant correlations were found between traits. As mentioned above for scions, traits related to genetic variability in rootstock exhibited a polygenic control as revealed by the detection of multiple QTLs. One QTL for water extraction capacity was identified in the three years on linkage group 3, confirming the hypothesis proposed by Carbonneau (1985) and Soar et al. (2006a) that this trait was genetically controlled at the rootstock level. A genetic architecture of transpiration plasticity to water deficit was evidenced which was partially independent from the genetic architecture of transpiration rate, suggesting an independent selection process for these two traits. Riparia Gloire de Montpellier, reputed as sensitive to water deficit, early reduced its scion transpiration as the soil was drying. The genetic architectures of δ13C and transpiration efficiency were partially independent, underlining the complexity of selecting plant material for water use efficiency (Condon et al., 2004). Transpiration efficiency appeared to be less influenced by climatic (year) effect and soil water conditions, and could therefore be more easily used for breeding. The QTLs detected in the offspring included genes that have been characterized as potentially involved in water deficit responses (Marguerit et al., 2012). Candidate genes related to hormone (notably ABA) and hydraulic (aquaporins) signaling between the rootstock and the scion are particularly interesting as they play a major role in water deficit responses (Soar et al., 2006a; Vandeleur et al., 2009).

This review and other data collected on rootstocks show that drought tolerance may probably be acquired through different mechanisms (Serra et al., 2014; Rossdeutsch et al., 2016). This diversity should be taken into account to adapt plant material to different situations and levels of water deficit.


Grapevine response to water limitation is complex and involves many physiological mechanisms. Genetic variability has been described for several traits related to these mechanisms and many associated genomic regions have already been identified at the scion and rootstock levels. Better knowledge on the role of favorable alleles in these regions will help designing adequate plant material to deal with the increased risk of drought events in the context of climate change.

Acknowledgments : This work was supported by funding from the project Long-Term Adaptation to Climate Change in Viticulture and Enology (LACCAVE) of the French National Institute for Agricultural Research (INRA).